A facial recognition match put an innocent woman in jail for six months — and the police department that ran it had no policy for using the technology at all. That gap is now a $10 million federal lawsuit, and it lands the same week OpenAI's chief executive heads to the UN Security Council to talk about AI risk.

Angela Lipps, 50, is suing the city of Fargo, North Dakota and one of its detectives for $10 million after an AI facial recognition match tied her to a string of bank thefts in a state she had never visited — and kept her in custody for six months. According to the complaint, a West Fargo investigator ran facial recognition on surveillance footage and the software returned Lipps's social media photo as a potential lead. The suspect in that footage had used a fake ID, and the complaint states Lipps did not resemble the woman in the video. Detective Lucas Heck is accused of telling a court that facial recognition had identified Lipps, then requesting an arrest warrant without further investigation or sufficient probable cause. Lipps was arrested at gunpoint at her Tennessee home, extradited, and jailed; the suit says she was denied her medications and her dentures while in custody, spent her birthday and the holidays behind bars, and came home to find her residence, car and belongings gone. Fargo declined to comment on pending litigation.

The claim that matters legally is not that the software was wrong — facial recognition is wrong all the time — but that the department knew it could be wrong and had no rules for it. The complaint alleges Fargo police lacked policy, training and supervision on facial recognition, and that its officers "knew that AI facial recognition technology should not be used as the sole basis to extradite and charge a person." That is the industry-standard rule: a facial recognition hit is an investigative lead, not probable cause, and it has to be corroborated before anyone is charged. The question a court will now test is whether the absence of that rule is negligence the city has to pay for. We covered how far the technology has already spread into ordinary policing — WA police facial recognition trial scanned 130,000 faces — and this case is the first serious bill for what happens when nobody writes the policy down.

Sam Altman will brief the UN Security Council next week on AI and international security, according to Reuters, which cited an OpenAI spokesperson. The appearance is scheduled around the UN General Assembly's annual gathering of world leaders in New York. It follows UN Secretary-General António Guterres's warning last week that the AI race could end in a "gigantic disaster" — a line we covered in UN chief warns the AI race could end in a 'gigantic disaster'.

A briefing is not governance. The Security Council's five veto-wielding members include the three countries that host the world's leading labs, and none of them has shown appetite for binding rules on their own industry. What Altman can offer that diplomats cannot is operational detail — what frontier systems actually do, where the failure modes are, how fast the capability curve is moving. What he cannot offer is enforcement, and the optics of a lab CEO explaining AI risk to the body that would have to regulate him are exactly the tension the trip will surface. Watch whether the session produces a resolution, a working group, or just a readout.

Should police departments be barred from making arrests on a facial recognition hit alone — or is a corroboration rule enough? Tell us in the comments.